Using doctest with pandas, how to fix the number of columns in the output example?

Question

I have a package with many methods that output pandas data frame. I would like to test the examples with pytest and doctest as explained on the pytest doctest integration page.

Pytest requires the output data frame to contain a certain number of columns that might be different than the number of columns provided in the example.

    >>> import pandas
    >>> df = pandas.DataFrame({"variable_1": range(3)})
    >>> for i in range(2, 8): df["variable_"+str(i)] = range(3)
    >>> df
       variable_1  variable_2  variable_3  variable_4  variable_5  variable_6  variable_7
    0           0           0           0           0           0           0           0
    1           1           1           1           1           1           1           1
    2           2           2           2           2           2           2           2

pytest --doctest-modules returns the following error because it displays 6 columns instead of 7

Differences (unified diff with -expected +actual):
    @@ -1,4 +1,6 @@
    -   variable_1  variable_2  variable_3  variable_4  variable_5  variable_6  variable_7
    -0           0           0           0           0           0           0           0
    -1           1           1           1           1           1           1           1
    -2           2           2           2           2           2           2           2
    +   variable_1  variable_2  variable_3  ...  variable_5  variable_6  variable_7
    +0           0           0           0  ...           0           0           0
    +1           1           1           1  ...           1           1           1
    +2           2           2           2  ...           2           2           2
    +<BLANKLINE>
    +[3 rows x 7 columns]

Is there a way to fix the number of column? Does doctest always have a fixed terminal output?